Error detection/correction and fault detection/recovery – Pulse or data error handling – Digital data error correction
Reexamination Certificate
1998-11-17
2002-03-05
Baker, Stephen M. (Department: 2133)
Error detection/correction and fault detection/recovery
Pulse or data error handling
Digital data error correction
C341S058000, C341S059000
Reexamination Certificate
active
06353912
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to an encoding circuit, an encoding method, a digital signal transmitting apparatus, and a digital signal recording/reproducing apparatus.
2. Description of the Related Art
Trellis coded partial response method is an effective method as a signal processing method so as to record a digital signal in a high density to some recording medium, such as a magnetic medium. The Trellis coded partial response method is a combination of a partial response and a maximum-likelihood decoding process, that is performed with a restriction condition of a modulation code.
Extended partial response class
4
(referred to as EPR
4
) has been conventionally used as a partial response. In EPR
4
, an equalized waveform of a di-pulse response is expressed as (1, 1, −1, −1) at a sample point (symbol point). The system polynomial of EPR
4
is expressed as Formula (1).
G
(
D
)=(1−
D
)(1+
D
)
2
(1)
where D is a one-bit delay operator.
Next, trellis coded extended partial response class
4
(hereinafter referred to as TCEPR
4
) will be described. TCEPR
4
is used to perform the maximum-likelihood decoding process using EPR
4
and a restriction condition of a modulation code.
FIG. 1
shows an example of the structure of a circuit that performs TCEPR
4
. Input data I to be recorded is supplied to an encoder
201
. The encoder
201
performs a predetermined encoding process for the input data I and outputs encoded data CO. The encoded data CO is supplied to a magnetic recording channel
202
.
A circuit that converts input data into binary data is disposed upstream of the encoder
201
. The input data I is binary data. The magnetic recording channel
202
contains a recording circuit, a recording magnetic head, a magnetic record medium, a reproducing head, and a reproducing circuit. The recording circuit processes the encoded data CO so as to record the input data I to the magnetic record medium. In other words, the magnetic recording channel
202
is a portion that writes/records data to/from the magnetic record medium.
An equalizer
203
equalizes a signal reproduced from the magnetic record medium through the magnetic recording channel
202
. An output signal of the equalizer
203
is supplied to a maximum-likelihood decoder
204
. The maximum-likelihood decoder
204
performs a maximum-likelihood decoding process for the output signal of the equalizer
203
. A decoder
205
disposed downstream of the maximum-likelihood decoder
204
finally reproduces information recorded on the magnetic record medium. In reality, after a digital process is performed for sampled data that has been A/D converted, a reproducing process is performed downstream of the magnetic recording channel
202
.
An output signal of the equalizer
203
(namely, a signal that has been reproduced and equalized by the magnetic recording channel
202
) (this output signal is hereinafter referred to as a reproduced/equalized signal) has five levels of (−2, −1, 0, +1, and +2).
FIG. 2
shows an example of the reproduced/equalized signal. To restore the reproduced/equalized signal into binary data (namely, the encoded data that has not been recorded/reproduced by the magnetic recording channel
202
), Viterbi decoding process that is one kind of maximum-likelihood decoding process is used.
In the Viterbi decoding process, the maximum-likelihood sequence (path) is estimated as reproduced encoded data based on a result of a calculating process with available sequences of sampled data. Thus, the Viterbi decoding process has a high detecting performance. However, depending on a sequence of the reproduced/equalized signal, the maximum-likelihood sequence may not be easily settled corresponding to the results of the above-described calculating process. In this case, the calculated results are stored in a memory of the Viterbi decoder until the maximum-likelihood sequence is settled. Thus, if the length of an unsettled sequence (namely, the length of a sequence that is calculated until the maximum-likelihood sequence is settled) exceeds the memory length, the memory overflows and thereby an error takes place.
As an example of which the memory overflows, a maximum-likelihood sequence estimated from some types of sequences of values of a reproduced/equalized signal is not eternally settled. Such a sequence is referred to as quasi catastrophic sequence. In addition, before a sequence is settled as the maximum-likelihood sequence, if the length of an unsettled sequence exceeds the memory length, the memory overflows. Thus, when the quasi catastrophic sequence is removed and the length of an unsettled sequence is limited to the memory length, the memory can be prevented from overflowing.
An object of the channel encoding process is to trim a signal spectrum and limit the number of successive “0s” so as to improve the accuracy for extracting clock information. Another object of the channel encoding process is to prevent the memory from overflowing.
In recording/reproducing mode, the channel encoding process is performed in the following manner. Binary data to be recorded (hereinafter referred to as information word) is converted into binary data (hereinafter referred to as record word) corresponding to a predetermined conversion rule. The record word is recorded/reproduced to/from the magnetic recording channel
202
. In the reproducing mode, data that has been decoded by the above-described maximum-likelihood decoding process is inversely converted and the original information word is reproduced. (When there is no encoded error, the data decoded by the maximum-likelihood decoding process matches the record word).
In
FIG. 1
, the information word is I and the record word is CO. The predetermined conversion rule is an encoding rule performed by the encoder
201
. In the reproducing system, the decoder
205
performs a decoding process as an inversely converting process for data that has been decoded by the maximum-likelihood decoder
204
and thereby reproduces the original information word.
On the other hand, since the system polynomial of TCEPR
4
is expressed as Formula (1), the transfer function of the TCEPR
4
channel is a spectrum of which the Nyquist frequency is null. When data is encoded, a code conversion is performed in such a manner that the frequency component of the Nyquist frequency in the power spectrum density of the record current of which “1” and “0” of a code word sequence are recorded corresponding to the direction of the current becomes null. At the frequency of which the frequency component of the transfer function of the channel is null, data is encoded in such a manner that the frequency component of the power spectrum density of the code word sequence becomes null. Thus, the signal detection gain of decoded data can be improved.
To prevent the memory from overflowing in the Viterbi decoding process and improve the signal detection gain, the following encoding process is performed. In other words, by limiting the range of the variation of ADS (Alternating Digital Sum), data can be encoded in such a manner that the Nyquist frequency component becomes null. In TCEPR
4
, a code word that satisfies the limitation of the range of the variation of the ADS is used.
Assuming that the total number of bits of the input binary data is denoted by n, the ADS of a sequence {a
1
, . . . , an} is expressed as Formula (2).
ADS
=
∑
i
=
1
n
(
-
1
)
(
2
)
For example, when data is encoded corresponding to an encoding state transition chart shown in
FIG. 3
, a code word sequence of which the range of the variation of the ADS is limited to up to eight is generated. Under such a limitation, the square of the minimum Euclidean distance between sequences that are output from the TCEPR
4
channel becomes six. Thus, a high signal detection gain can be obtained.
FIG. 4
shows an example of a trellis available with a code word of for example 10 bits in the case that data is encoded in the cond
Baker Stephen M.
Frommer William S.
Frommer & Lawrence & Haug LLP
Kessler Gordon
LandOfFree
Encoding circuit, encoding method, digital signal... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Encoding circuit, encoding method, digital signal..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Encoding circuit, encoding method, digital signal... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2837368